In the fall of 2009, a large utility company in the American Midwest offered its employees free flu shots at clinics in their own workplaces. Clinics ran at 62 locations between October 13 and November 16. Some lasted part of a single day; others ran for five.
In early October, a healthcare communications firm mailed a reminder to every employee the flu vaccine was recommended for: people aged 50 or older and people with chronic conditions that raised the risk of complications. There were 3,272 of them.
Every letter gave the times and places of the clinics and explained how and why to get the shot. A team of behavioral researchers led by Katherine Milkman had arranged for three versions, assigned at random within each work location. At sites where the clinic ran for only one day, a date plan made no sense, so employees there got either the control letter or the date-and-time version.
The control letter stopped there. The other two added one line: “Many people find it helpful to make a plan for getting their shot. You can write yours here:”. Below it, one version left space for a date. The other left space for a date and a time.
Nobody collected the plans. The researchers never learned who wrote anything down, or whether anyone read the letter at all. The employer didn’t see the plans either. What the researchers measured was who got a shot at a workplace clinic, from administrative records.
In the control group, 33.1 percent of employees got vaccinated at a workplace clinic. Among those whose letter asked for a date, 35.6 percent did, a difference the study could not distinguish from chance. Among those asked for a date and a time, 37.1 percent did, an increase of 4.0 percentage points, statistically significant.
The prompt cost nothing extra to print or mail. Its effect was largest where the chance was shortest. At locations with a single-day clinic, the date-and-time letter raised vaccination by 7.9 to 9.5 percentage points, depending on the statistical model. At multi-day sites, the increase was smaller and not statistically significant. The researchers hadn’t predicted that split, and warned it might reflect other differences between the workplaces.
They listed other limits too. The employees, mostly men, all worked for one firm. And the fall of 2009 was unusual: news about the H1N1 swine flu and shortages of its separate vaccine was everywhere.
The study appeared in the Proceedings of the National Academy of Sciences in June 2011. Most of the employees who were asked for a date and a time still didn’t come. Two blanks had moved about four in every hundred.